Xirp is powered by Portal. By joining the beta you’ll get immediate access to Portal.
AI coding tools solved the generation problem. Code ships faster, more services get built, output goes up.
But something else happened. It took new engineers months to ramp up. AI agents made fast, confident decisions that were technically correct and operationally wrong — because they didn't know what they were actually working inside.
The knowledge that would fix this wasn't missing. It was in Slack threads nobody could find, in the heads of the three people who built that service in 2021.
That's not a documentation problem. It's a retrieval problem. And wikis, Confluence pages, and architecture diagrams don't solve it — they just go stale while the real work keeps moving.
Xirp is different. It's an AI coding agent that connects to your systems through Portal. Your services, ownership, dependencies, architectural decisions — and uses that context in every session. Your agents don't guess. They know.
An AI coding agent with institutional memory.
Other agents see the file you're working in. Xirp sees the system it's part of: who owns the upstream service, what depends on it, why it was built this way. Every session starts grounded, and every session makes the next one smarter.
The context layer your agents and engineers share.
Workspace, a Portal plugin, holds your work items, sessions and docs in one place. When an engineer wraps a session, the context doesn't disappear. It's stored, searchable and ready for the next person, or agent, who touches that system.
Living documentation, built automatically.
Every coding session generates real knowledge. Xirp captures it, turns it into documentation that stays current, and feeds it back into future sessions so every agent and engineer starts smarter. No more asking engineers to stop and write docs. It happens as the work happens.
Works with Claude, Gemini, Codex. Knows your org, not theirs.
Xirp is a harness for the model you prefer. Switch between Claude, Gemini and Codex without losing context. Run sessions locally or remotely. Stay vendor-independent while the context layer compounds with every session.
